The ISAC β-NMR HomePage

Introduction to β-NMR
Students Projects
Technical Information
Data Acquistion
Approved Experiments
Internal Page
Related Links
Data Analysis
Photo Album

  • Download BNMRfit istallation script
  • Download BNMRfit
  • Download BNMRminfit

  • The Polarizer

    Optical Polarization for β-NMR at ISAC

    In conventional NMR experiments, nuclear spins are polarized by a large static magnetic field. In the β-NMR apparatus at ISAC nuclear polarization is produced in the ion beam in flight via optical pumping with circularly polarized laser light. The polarizer relies on the well known technique, developed for many years at CERN's ISOLDE, of collinear optical pumping to highly polarize the nuclear spins of 10–60 keV radioactive beams. At ISAC, alkali-metal beams are longitudinally polarized through optical pumping of a fast atomic beam, which is created with up to 90% efficiency by charge exchange of the incident ion beam in a Na vapour jet (see Fig. 1). The polarized beam is then reionized in a cold He gas target with 66% efficiency and directed to the experiments.

    Fig. 1. Details of the collinear optically pumped polarizer.

    Reionizing the beam gives some important advantages, enabling the use of electrostatic optics to steer and focus the beam into several experiments. Two or more experiments may run simultaneously in the future by sharing the beam using a fast kicker. The beam energy, and hence implantation depth, is varied at the existing condensed matter experimental station from a few hundred eV to 60 keV, by biasing the target. Post-polarization electrostatic bends isolate the experiments from the Na vapour and laser light in the polarizer. The polarization direction (longitudinal or transverse with respect to the beam momentum) is determined by the number of electrostatic 90° bends. A polarimeter measures the transverse polarization of the ion beam after it has been electrostatically deflected by 90°, and a second polarimeter is being built to measure the longitudinal polarization of the undeflected fast atomic beam. To date, 30 keV polarized beams of 8Li+ (over 107 ions s-1 at the experiment) and 9Li+ (105 ions s-1) have been produced at ISAC.

    The 8Li beam is polarized by optical pumping on the D1 transition with counter-propagating circularly polarized light. 8Li has two ground state hyperfine levels, in common with most alkali-metal isotopes, that in this case are split by 382 MHz (see Fig. 2). Both levels must be pumped in order to achieve high polarization. The optical pumping laser is an argon-ion laser pumped, standing wave Spectra-Physics 3900S Ti:sapphire (TiS) laser, specially modified to lase at the two required frequencies at 673 nm. This is very near the short wavelength limit of TiS, and is achieved with a rear cavity mirror centred at 650 nm and an R=99% output coupler. The cavity length is shortened to increase the longitudinal mode spacing to 381 MHz. A 1 mm thick, R=10% intra-cavity etalon ensures operation on only two neighboring modes. The crystal gain medium is in the centre of the cavity, where the nodes of one mode coincide with the anti-nodes of the other, minimizing gain competition and ensuring simultaneous oscillation of both modes. Typical laser power is ~100 mW on each mode. The linewidth of each mode is ~1 MHz on timescales comparable to the 2 small mu, Greeks transit time of the atoms in the optical pumping region. Vibrational instability causes frequency fluctuations over a range of ±20 MHz on much longer timescales. Doppler-shift tuning onto resonance is done by varying the deceleration potential on the Na cell between 0 and 1 kV, thus scanning the beam energy, while keeping the laser frequencies fixed. The tuning peak is found by measuring the polarization at the polarimeter, or observing the laser induced fluorescence in the optical pumping region. The laser frequencies are prevented from drifting more than ±3 MHz per day by a system based on a frequency stabilized He–Ne laser.


    Fig. 2. A scheme of the atomic levels of 8Li. The upper part of the figure shows the hyperfine splittings in the ground state and 1st excited state of 8Li, along with the pumping transitions. The lower part shows the nearly degenerate Zeeman splitting of the hyperfine levels into 2F+1 magnetic sublevels m_F. Optical pumping with sigma+ light is shown, in which the only allowed absorptions are those having an increase of angular momentum +1, Delta m_F = +1. Fluorescence can occur on transitions satisfying Delta m_F = 0, +/-1. After  many cycles of optical pumping, all the atoms end up in the fully stretched state m_F = 5/2, with full nuclear and electronic polarization, shown by the red circle. Once in that state, they cannot be pumped out of it by the laser light. The  luorescence lifetime is 27 ns, which allows for approximately 20 optical pumping cycles in the 2 us transit time of atoms through the polarizer. Pumping with sigma- light pumps all the atoms into the m_F = -5/2 ground state.

    The key to achieving high polarization is to match the optical pumping light bandwidth to the energy (Doppler) broadening of the Li beam. The typical energy broadening is about 100 MHz, caused by multiple collisions with Na atoms in the neutralizer, and cannot be avoided if high neutralization efficiency is required. To overcome this problem, we use resonant electro-optic modulators (EOMs) to produce laser sidebands and broaden the effective laser bandwidth. Rate equation calculations showed that a 19 MHz EOM and a 28 MHz EOM would work effectively in series, in terms of both the total width and the sideband spacing of about 10 MHz. Cesium vapour was tested as a neutralizer, since it was thought that its smaller excitation energy than Na and greater mass would produce less energy broadening. However, the energy broadening at useful Cs densities was found to be larger than in Na vapour.

    The Polarization (reflected in the detected beta decay asymmetry) as the laser doublet is swept through the atomic resonance


    Fig. 3. Typical tuning signal, showing β-decay asymmetry (uncorrected for different detector efficiencies) for both light helicities.The statistical error bars are too small to show. The variable offset is due to the beam shifting position on the foil.

    For more information can be found in  Atsushi's ISAC Polarizer Page. The idea of optical pumping is quite old, and it has been applied in several ways, here is a brief bibliography:

    Preliminary work:

    1. W. Hanle, Z. Phys. 30, 93 (1924).

    Original suggestion and demonstrations:

    1. A. Kastler, J. Phys. et le Radium 11, 233 (1950).
    2. J. Brossel, A. Kastler and J. Winter, J. Phys. et le Radium 13, 668 (1952).
    3. W.B. Hawkins and R.H. Dicke, Phys. Rev. 91, 1008 (1953).

    In semiconductors:

    1. G. Lampel, Phys. Rev. Lett. 20, 491 (1968). 
    2. S.E. Barrett, R. Tycko, L.N. Pfeiffer and K.W. West, Phys. Rev. Lett. 72, 1368 (1994).

    Page last modified: 07/23/09 02:45 by Andrew MacFarlane.